Unveiling the Dynamics of Pedestrian Movement in Center City, Philadelphia

Xiayuanshan Gao

Overview

In this comprehensive study of pedestrian movement in Center City, Philadelphia, I’ve dissected walking patterns through a multi-faceted lens to understand the underpinnings of foot traffic dynamics in 2012. Initially, I examined the types of walking trips and the socioeconomic profiles of pedestrians, revealing a preference for walking across a spectrum of income brackets, age groups, and educational levels. Subsequently, I mapped the geography of pedestrian travel, highlighting the central district as the focal point of walking activities, enriched by its transit hubs and amenities. Delving into what attracts people to walk, I uncovered that restaurants, schools, commercial blocks, and green spaces play pivotal roles in drawing pedestrians. Lastly, I employed a regression model to quantify the influence of urban features on pedestrian movement, confirming the importance of amenities like restaurants and public transit in promoting walkability.

1. Profiling the Walking Trips and Pedestrians in Philadelphia in 2012

In the first section, I delve into the pedestrian dynamics of Philadelphia in 2012, aiming to unpack the walking trip patterns and profile those who most frequently travel by foot. My exploration spans various dimensions, from the origins and destinations of these journeys to the walkers’ socioeconomic characteristics, including income levels, vehicle ownership, gender, age, race, and education. This comprehensive overview not only illuminates the prevalent trends in walking trips but also highlights the diverse tapestry of individuals who opt for pedestrian travel, thereby offering an encompassing snapshot of urban mobility within the city.

import pandas as pd
import numpy as np
import altair as alt
import geopandas as gpd
import hvplot.pandas
import numpy as np
import pandas as pd
import seaborn as sns
from matplotlib import pyplot as plt
import datetime
import math
from pandana.loaders import osm
import folium

trip = pd.read_excel('data/4_Trip_Public.xlsx')
trip.head ()

	ID	HH_ID	PERSON_NUM	PERSON_ID	HH_WEIGHT	P_WEIGHT	TOUR_NUM	WKSUB_NUM	WKSUB_ID	TRIP_NUM	...	D_COUNTY	D_CPA	D_MCD	D_TRACT	D_TAZ	GPSFactor	Survey_TravTime	Model_TravTime	Model_TravDist	CompositeWeight
0	1	100140	1	10014001	133.460324	132.54307	1.0	NaN	1.001400e+11	1	...	34007.0	007_802	3.400712e+09	3.400760e+10	22228.0	1.12	15.0	7.456654	1.535654	148.448239
1	2	100140	1	10014001	133.460324	132.54307	1.0	NaN	1.001400e+11	2	...	34007.0	007_802	3.400712e+09	3.400760e+10	22231.0	1.12	7.0	7.810088	1.675524	148.448239
2	3	100140	1	10014001	133.460324	132.54307	1.0	NaN	1.001400e+11	3	...	34005.0	005_702	3.400544e+09	3.400570e+10	20225.0	1.12	12.0	9.363016	2.301180	148.448239
3	4	100140	1	10014001	133.460324	132.54307	1.0	NaN	1.001400e+11	4	...	34005.0	005_702	3.400544e+09	3.400570e+10	20227.0	1.12	16.0	6.233144	1.086251	148.448239
4	5	100140	1	10014001	133.460324	132.54307	1.0	NaN	1.001400e+11	5	...	34005.0	005_702	3.400544e+09	3.400570e+10	20225.0	1.12	10.0	6.258231	1.208349	148.448239

5 rows × 82 columns

# Trim to Phily and bike trip
walk = trip[(trip['O_COUNTY'] == 42101) & (trip['MODE_AGG'] == 1)]

walk.head()

	ID	HH_ID	PERSON_NUM	PERSON_ID	HH_WEIGHT	P_WEIGHT	TOUR_NUM	WKSUB_NUM	WKSUB_ID	TRIP_NUM	...	D_COUNTY	D_CPA	D_MCD	D_TRACT	D_TAZ	GPSFactor	Survey_TravTime	Model_TravTime	Model_TravDist	CompositeWeight
12	13	100206	1	10020601	204.106167	281.143358	1.0	NaN	1.002060e+11	3	...	42101.0	101_103	4.210160e+09	4.210100e+10	113.0	1.17	15.0	16.783789	0.867162	328.937729
13	14	100206	1	10020601	204.106167	281.143358	1.0	NaN	1.002060e+11	4	...	42101.0	101_103	4.210160e+09	4.210100e+10	44.0	1.17	11.0	12.350352	0.638102	328.937729
302	314	102372	1	10237201	209.139870	218.215946	1.0	NaN	1.023720e+11	4	...	42101.0	101_103	4.210160e+09	4.210100e+10	35.0	1.13	9.0	NaN	NaN	246.584019
303	315	102372	1	10237201	209.139870	218.215946	1.0	NaN	1.023720e+11	5	...	42101.0	101_103	4.210160e+09	4.210100e+10	38.0	1.13	10.0	11.381527	0.588046	246.584019
991	1009	109412	1	10941201	193.156252	82.416676	1.0	NaN	1.094120e+11	3	...	42101.0	101_103	4.210160e+09	4.210100e+10	55.0	1.13	10.0	16.072276	0.830401	93.130844

5 rows × 82 columns

1.1 What Type of Walking Trips Happened in Philadelphia in 2012

From the presented data, it is evident that walking trips in Philadelphia in 2012 were predominantly initiated and concluded at a variety of locations classified under ‘Other’, denoting a diverse range of non-home and non-work-related starting and ending points. This diversity in trip locations correlates with a higher frequency of walking trips associated with home activities that are not related to work, school, or online engagements. Despite the variety in origins and destinations, there is a notable preference for walking trips that end at ‘Home’, suggesting that Philadelphia residents may prefer walking for returning home from various activities. In contrast, walking trips starting from ‘Work’ are less frequent than those ending there, hinting at the possibility that walking is less preferred for commuting to work. The subsequent chart detailing the types of activities associated with walking trips reinforces the inference that non-work-related activities, particularly those centered around the home, are the primary motivators for walking.

The first pair of charts compare the frequency of different types of locations from where people start and end their walking trips. The categories include ‘Other’, ‘Work’, ‘Home’, and ‘School’. For origins, the ‘Other’ category has the highest count, indicating that most walking trips begin at locations other than work, home, or school. This is followed by ‘Work’, ‘Home’, and ‘School’ which has the least count. For destinations, ‘Other’ also has the highest count, suggesting that destinations are similarly varied. However, ‘Home’ has a significantly higher count as a destination than as an origin, while ‘Work’ has a lower count as a destination than as an origin. ‘School’ remains the least frequent destination.

replacement_dict = {1: 'Home', 2: 'Work', 3: 'School', 4: 'Other'}

walk['O_LOC_TYPE'] = walk['O_LOC_TYPE'].replace(replacement_dict)
walk['D_LOC_TYPE'] = walk['D_LOC_TYPE'].replace(replacement_dict)

sns.set_style("whitegrid")

fig, ax = plt.subplots(1, 2, figsize=(12, 6))

# Plot Origin Loc Type 
sns.countplot(x='O_LOC_TYPE', data=walk, ax=ax[0])
ax[0].set_title('Frequency of  Origin Loc Type ')
ax[0].set_xlabel('Origin Loc Type ')
ax[0].set_ylabel('Count')

# Plot Destination Loc Type 
sns.countplot(x='D_LOC_TYPE', data=walk, ax=ax[1])
ax[1].set_title('Frequency of Destination Loc Type')
ax[1].set_xlabel('Destination Loc Type')
ax[1].set_ylabel('Count')

# Show the plot
plt.tight_layout()
plt.show()

C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\1475632616.py:3: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  walk['O_LOC_TYPE'] = walk['O_LOC_TYPE'].replace(replacement_dict)
C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\1475632616.py:4: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  walk['D_LOC_TYPE'] = walk['D_LOC_TYPE'].replace(replacement_dict)

The following chart shows the frequency of different types of activities for which walking is used. The most common activity is ‘Home activities not related to work, school, or online’, indicating that walking is most frequently used for local errands or tasks around the house. This is followed by ‘Work for pay’ and ‘Change type of transportation/transfer’. Activities like ‘Social (visit friends, relatives, etc.)’, ‘Medical (medical appointment, medical procedure, etc.)’, and ‘Recreation-watch/observe (movies, concert, sports event, etc.)’ have lower frequencies.

# Find out activity type
activity_dict = { 1: 'Home activities not related to work, school, or online',
                 2: 'Homework, class related assignments or attended an online course',
                 3: 'Attended classes',
                 4: 'Attended other school activities (performances, meetings, clubs)',
                 5: 'Work for pay',
                 6: 'Personal business (banking or ATM, salon, library)',
                 7: 'Online personal business (banking, e-mail, etc.)',
                 8: 'Everyday shopping (grocery, drug store, gas, etc.)',
                 9: 'Major shopping (appliances, cars, home furnishings, clothes, etc.)',
                 10: 'Online shopping for products, services or goods',
                 11: 'Eat out (restaurant, drive-thru, etc.)',
                 12: 'Social (visit friends, relatives, etc.)',
                 13: 'Social/community/religious (meetings, worhship, wedding, funeral, etc.)',
                 14: 'Recreation- active participation (sports, exercise, walk the dog, etc.)',
                 15: 'Recreation-watch/observe (movies, concert, sports event, etc.)',
                 16: 'Medical (medical appointment, medical procedure, etc.)',
                 17: 'Pick up passenger',
                 18: 'Drop off passenger',
                 19: 'Change type of transportation/transfer',
                 20: 'Accompany household member',
                 96: 'Other activity'}

walk['ACTIV1'] = walk['ACTIV1'].replace(activity_dict)

plt.figure(figsize=(10, 8))


activity_counts = walk['ACTIV1'].value_counts()
sns.barplot(y=activity_counts.index, x=activity_counts.values)
plt.title('Frequency of Walk Activity Type')
plt.xlabel('Count')
plt.ylabel('Activity')

for index, value in enumerate(activity_counts.values):
    plt.text(value, index, f'{value}')

plt.show()

C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\28425172.py:24: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  walk['ACTIV1'] = walk['ACTIV1'].replace(activity_dict)

1.2 Who Prefer to Walk

Then, the following analysis of Philadelphia’s pedestrian demographics in 2012 reveals a nuanced portrait influenced by various socioeconomic factors. The data on total vehicles indicates that individuals with no vehicles in their household are the most likely to walk, which is consistent across all income brackets except for the highest and lowest, where walking is less prevalent. Income distribution among pedestrians shows a fairly even spread, with slight peaks in the middle-income ranges, suggesting that walking is not exclusively linked to any single income level.

Gender distribution shows a modest preference for walking among females compared to males. Age-wise, the 25-74 age group is the most inclined to walk, with a notable peak in the 55-59 bracket, highlighting the potential for walking as a consistent transportation choice across a broad adult age spectrum.

Racial demographics indicate that White individuals and those identifying as Asian are more represented among walkers than other races, with Black/African American and Hispanic or Latino individuals walking less by comparison. Education level data shows a clear trend: individuals with higher educational attainment, specifically those with a bachelor’s or graduate degree, are more likely to walk than those with lower levels of education or some college credit.

In summary, the typical pedestrian in Philadelphia in 2012 would likely be a person without a vehicle in their household, belonging to the middle-income bracket, and leaning slightly towards the female demographic. Age-wise, adults, particularly those between 25 to 74 years old, with a concentration around late 50s, are the most common walkers. A higher educational background correlates positively with the likelihood of walking, while White and Asian residents tend to walk more than other racial groups. These insights draw a picture of pedestrians in Philadelphia as a diverse group with a significant representation among middle-aged, educated individuals, suggesting that walking is a transportation mode that cuts across various socioeconomic layers, albeit with certain demographic tendencies.

# Load other df
hh = pd.read_excel('data/1_Household_Public.xlsx')

per = pd.read_excel('data/2_Person_Public.xlsx')

import pandas as pd
import matplotlib.pyplot as plt

walk = walk.merge(hh[['HH_ID', 'TOT_VEH', 'INCOME']], on='HH_ID', how='left')
walk = walk.merge(per[['PERSON_ID', 'GEND', 'AGECAT', 'RACE', 'EDUCA']], on='PERSON_ID', how='left')

# Create mappings for the variables hh and person 
gend_mapping = {1: 'Male', 2: 'Female', 99: 'Refused'}
agecat_mapping = {
    1: '5 or younger', 2: '6 to 12', 3: '13 to 15', 4: '16 to 17', 
    5: '18 to 24', 6: '25 to 34', 7: '35 to 44', 8: '45 to 54', 
    9: '55 to 64', 10: '65 to 74', 11: '75 to 85', 12: '86 and over',
    98: "Don't know", 99: 'Refused'
}

race_mapping = {
    1: 'White/Caucasian', 2: 'Black/African American', 3: 'Hispanic or Latino',
    4: 'American Indian or Alaska Native', 5: 'Asian', 6: 'Native Hawaiian or Pacific Islander',
    97: 'Other, Specify', 98: "Don't Know", 99: 'Refused', 100: 'Multi-race'
}
educa_mapping = {
    1: 'Not a high school graduate', 2: 'High school graduate', 3: 'Some college credit',
    4: 'Associate or technical school degree', 5: "Bachelor's or undergraduate degree",
    6: 'Graduate degree', 97: 'Other', 98: "Don't know", 99: 'Refused'
}
income_mapping = {
    1: '$0 to $9,999', 2: '$10,000 to $24,999', 3: '$25,000 to $34,999', 4: '$35,000 to $49,999',
    5: '$50,000 to $74,999', 6: '$75,000 to $99,999', 7: '$100,000 to $149,999',
    8: '$150,000 to $199,999', 9: '$200,000 to $249,999', 10: '$250,000 or more',
    98: "Don't know", 99: 'Refused'
}

walk['GEND'] = walk['GEND'].map(gend_mapping)
walk['AGECAT'] = walk['AGECAT'].map(agecat_mapping)
walk['RACE'] = walk['RACE'].map(race_mapping)
walk['EDUCA'] = walk['EDUCA'].map(educa_mapping)
walk['INCOME'] = walk['INCOME'].map(income_mapping)

fig, axs = plt.subplots(2, 3, figsize=(15, 10))  

walk['TOT_VEH'].value_counts().sort_index().plot(kind='bar', ax=axs[0, 0], title='Total Vehicles')
walk['INCOME'].value_counts().sort_index().plot(kind='bar', ax=axs[0, 1], title='Income')
walk['GEND'].value_counts().sort_index().plot(kind='bar', ax=axs[0, 2], title='Gender')
walk['AGECAT'].value_counts().sort_index().plot(kind='bar', ax=axs[1, 0], title='Age Category')
walk['RACE'].value_counts().sort_index().plot(kind='bar', ax=axs[1, 1], title='Race')
walk['EDUCA'].value_counts().sort_index().plot(kind='bar', ax=axs[1, 2], title='Education')

plt.tight_layout()
plt.show()

In conclusion, the multifaceted examination of walking trips in Philadelphia in 2012 reveals that walking is a mode of transportation embraced across a spectrum of socioeconomic backgrounds, with particular tendencies shaped by the absence of vehicles, middle-income brackets, and higher educational attainments. Notably, the proclivity to walk is slightly more common among females and notably prevalent in the 25-74 age demographic, with a surprising peak among those in their late fifties. Additionally, while White and Asian residents are more represented among pedestrians, the data indicates that walking is a universally adopted practice with a presence in every income category and demographic segment. These findings reflect the inherent complexity of pedestrian movement patterns and suggest that walking, as an integral element of urban life, is influenced by a confluence of personal, economic, and cultural factors that define the pedestrian landscape of Philadelphia.

2. Where Do People Walk in Philadelphia in 2012

In the second section of my analysis, I turn my focus to the spatial dynamics of pedestrian movement within Philadelphia, unraveling the patterns of where residents choose to walk. Through a series of flow maps and heat maps, I discern the urban fabric’s influence on pedestrian behavior. The central district emerges as the epicenter of pedestrian activity, with a vibrant flux of walkers traversing its streets, likely due to commercial, cultural, and civic amenities. Adjacent areas such as the lower north, south, University southwest, and west districts also exhibit significant pedestrian traffic, albeit less densely clustered than the central district. In stark contrast, the lower southwest and lower south districts present as pedestrian deserts, areas where foot traffic is markedly sparse. These regions may lack pedestrian-friendly infrastructure or destinations that encourage walking. The overlaid heat maps, distinguishing between origins and destinations, further underscore these patterns, revealing hotspots of pedestrian origination and culmination. When combined, these visual tools paint a detailed picture of walking patterns across the city, highlighting areas of high pedestrian engagement and pinpointing zones that could benefit from urban design interventions to enhance walkability and connectivity.

# load Census Tract

import pygris

pa_state_code = '42' 
philly_county_code = '101'

philly_tract = pygris.tracts(state=pa_state_code, county=philly_county_code, year=2012)

philly_tract.head(5)

	STATEFP	COUNTYFP	TRACTCE	GEOID	NAME	NAMELSAD	MTFCC	FUNCSTAT	ALAND	AWATER	INTPTLAT	INTPTLON	geometry
1783	42	101	038100	42101038100	381	Census Tract 381	G5020	S	3335030	1629584	+40.0189334	-075.0399564	POLYGON ((-75.06785 40.01121, -75.06752 40.011...
1784	42	101	038400	42101038400	384	Census Tract 384	G5020	S	2546786	0	+40.0714163	-075.2332568	POLYGON ((-75.24321 40.07356, -75.24313 40.073...
1785	42	101	038600	42101038600	386	Census Tract 386	G5020	S	3304398	32066	+40.0582762	-075.2110913	POLYGON ((-75.22199 40.07306, -75.22128 40.073...
1786	42	101	038500	42101038500	385	Census Tract 385	G5020	S	1338942	2576	+40.0757528	-075.2146234	POLYGON ((-75.22784 40.08423, -75.22774 40.084...
1787	42	101	038700	42101038700	387	Census Tract 387	G5020	S	2053212	22241	+40.0833122	-075.2130770	POLYGON ((-75.22977 40.08676, -75.22961 40.086...

philly_tract = philly_tract.to_crs(epsg=4326)

# Calculate the centroid points in geographic coordinates
philly_tract['centroid'] = philly_tract.geometry.centroid
philly_tract['centroid_lat'] = philly_tract.centroid.y
philly_tract['centroid_lon'] = philly_tract.centroid.x

philly_tract.head()

C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\3754762874.py:4: UserWarning: Geometry is in a geographic CRS. Results from 'centroid' are likely incorrect. Use 'GeoSeries.to_crs()' to re-project geometries to a projected CRS before this operation.

  philly_tract['centroid'] = philly_tract.geometry.centroid
C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\3754762874.py:5: UserWarning: Geometry is in a geographic CRS. Results from 'centroid' are likely incorrect. Use 'GeoSeries.to_crs()' to re-project geometries to a projected CRS before this operation.

  philly_tract['centroid_lat'] = philly_tract.centroid.y
C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\3754762874.py:6: UserWarning: Geometry is in a geographic CRS. Results from 'centroid' are likely incorrect. Use 'GeoSeries.to_crs()' to re-project geometries to a projected CRS before this operation.

  philly_tract['centroid_lon'] = philly_tract.centroid.x

	STATEFP	COUNTYFP	TRACTCE	GEOID	NAME	NAMELSAD	MTFCC	FUNCSTAT	ALAND	AWATER	INTPTLAT	INTPTLON	geometry	centroid	centroid_lat	centroid_lon
1783	42	101	038100	42101038100	381	Census Tract 381	G5020	S	3335030	1629584	+40.0189334	-075.0399564	POLYGON ((-75.06785 40.01121, -75.06752 40.011...	POINT (-75.03963 40.01770)	40.017698	-75.039634
1784	42	101	038400	42101038400	384	Census Tract 384	G5020	S	2546786	0	+40.0714163	-075.2332568	POLYGON ((-75.24321 40.07356, -75.24313 40.073...	POINT (-75.23326 40.07142)	40.071418	-75.233255
1785	42	101	038600	42101038600	386	Census Tract 386	G5020	S	3304398	32066	+40.0582762	-075.2110913	POLYGON ((-75.22199 40.07306, -75.22128 40.073...	POINT (-75.21027 40.05805)	40.058053	-75.210272
1786	42	101	038500	42101038500	385	Census Tract 385	G5020	S	1338942	2576	+40.0757528	-075.2146234	POLYGON ((-75.22784 40.08423, -75.22773 40.084...	POINT (-75.21369 40.07591)	40.075907	-75.213688
1787	42	101	038700	42101038700	387	Census Tract 387	G5020	S	2053212	22241	+40.0833122	-075.2130770	POLYGON ((-75.22977 40.08676, -75.22961 40.086...	POINT (-75.21434 40.08330)	40.083299	-75.214342

philly_tract['TRACTCE'] = philly_tract['TRACTCE'].astype(str)

walk['O_TRACT'] = walk['O_TRACT'].astype(str)
walk['D_TRACT'] = walk['D_TRACT'].astype(str)

# only keep the strings for tract
walk['O_TRACT'] = walk['O_TRACT'].str[5:11]
walk['D_TRACT'] = walk['D_TRACT'].str[5:11]

walk.head()

	ID	HH_ID	PERSON_NUM	PERSON_ID	HH_WEIGHT	P_WEIGHT	TOUR_NUM	WKSUB_NUM	WKSUB_ID	TRIP_NUM	...	Survey_TravTime	Model_TravTime	Model_TravDist	CompositeWeight	TOT_VEH	INCOME	GEND	AGECAT	RACE	EDUCA
0	13	100206	1	10020601	204.106167	281.143358	1.0	NaN	1.002060e+11	3	...	15.0	16.783789	0.867162	328.937729	2	$50,000 to $74,999	Male	45 to 54	NaN	Bachelor's or undergraduate degree
1	14	100206	1	10020601	204.106167	281.143358	1.0	NaN	1.002060e+11	4	...	11.0	12.350352	0.638102	328.937729	2	$50,000 to $74,999	Male	45 to 54	NaN	Bachelor's or undergraduate degree
2	314	102372	1	10237201	209.139870	218.215946	1.0	NaN	1.023720e+11	4	...	9.0	NaN	NaN	246.584019	2	Refused	Male	35 to 44	NaN	Graduate degree
3	315	102372	1	10237201	209.139870	218.215946	1.0	NaN	1.023720e+11	5	...	10.0	11.381527	0.588046	246.584019	2	Refused	Male	35 to 44	NaN	Graduate degree
4	1009	109412	1	10941201	193.156252	82.416676	1.0	NaN	1.094120e+11	3	...	10.0	16.072276	0.830401	93.130844	2	Refused	Female	65 to 74	NaN	Graduate degree

5 rows × 88 columns

# Merge latitude and longitude for origins
walk = walk.merge(
    philly_tract[['TRACTCE', 'centroid_lat', 'centroid_lon']].rename(
        columns={'centroid_lat': 'origin_lat', 'centroid_lon': 'origin_lon','TRACTCE': 'origin_tra'}
    ),
    how='left', left_on='O_TRACT', right_on='origin_tra'
)

# Merge latitude and longitude for destinations
walk = walk.merge(
    philly_tract[['TRACTCE', 'centroid_lat', 'centroid_lon']].rename(
        columns={'centroid_lat': 'destination_lat', 'centroid_lon': 'destination_lon','TRACTCE': 'destination_tra'}
    ),
    how='left', left_on='D_TRACT', right_on='destination_tra'
)


walk.head()

	ID	HH_ID	PERSON_NUM	PERSON_ID	HH_WEIGHT	P_WEIGHT	TOUR_NUM	WKSUB_NUM	WKSUB_ID	TRIP_NUM	...	GEND	AGECAT	RACE	EDUCA	origin_tra	origin_lat	origin_lon	destination_tra	destination_lat	destination_lon
0	13	100206	1	10020601	204.106167	281.143358	1.0	NaN	1.002060e+11	3	...	Male	45 to 54	NaN	Bachelor's or undergraduate degree	000804	39.948566	-75.167755	000300	39.957150	-75.171232
1	14	100206	1	10020601	204.106167	281.143358	1.0	NaN	1.002060e+11	4	...	Male	45 to 54	NaN	Bachelor's or undergraduate degree	000300	39.957150	-75.171232	000804	39.948566	-75.167755
2	314	102372	1	10237201	209.139870	218.215946	1.0	NaN	1.023720e+11	4	...	Male	35 to 44	NaN	Graduate degree	001002	39.945006	-75.146618	001002	39.945006	-75.146618
3	315	102372	1	10237201	209.139870	218.215946	1.0	NaN	1.023720e+11	5	...	Male	35 to 44	NaN	Graduate degree	001002	39.945006	-75.146618	000902	39.947161	-75.156637
4	1009	109412	1	10941201	193.156252	82.416676	1.0	NaN	1.094120e+11	3	...	Female	65 to 74	NaN	Graduate degree	000500	39.951955	-75.158176	000600	39.949386	-75.158334

5 rows × 94 columns

import folium

average_lat = walk[['origin_lat', 'destination_lat']].stack().mean()
average_lon = walk[['origin_lon', 'destination_lon']].stack().mean()

od = folium.Map(location=[average_lat, average_lon], zoom_start=12,
    tiles='CartoDB positron')

for idx, row in walk.iterrows():
    if pd.notnull(row['origin_lat']) and pd.notnull(row['origin_lon']):
        folium.CircleMarker(
            location=[row['origin_lat'], row['origin_lon']],
            radius=3,
            color='skyblue',
            fill=True,
            fill_color='skyblue'
        ).add_to(od)

    if pd.notnull(row['destination_lat']) and pd.notnull(row['destination_lon']):
        folium.CircleMarker(
            location=[row['destination_lat'], row['destination_lon']],
            radius=3,
            color='pink',
            fill=True,
            fill_color='pink'
        ).add_to(od)
    
    if pd.notnull(row['origin_lat']) and pd.notnull(row['origin_lon']) and \
       pd.notnull(row['destination_lat']) and pd.notnull(row['destination_lon']):
        folium.PolyLine(
            locations=[
                [row['origin_lat'], row['origin_lon']],
                [row['destination_lat'], row['destination_lon']]
            ],
            color='lavender',
            weight=1
        ).add_to(od)

od

Make this Notebook Trusted to load map: File -> Trust Notebook

# Load the district boundaries file
districts_gdf = gpd.read_file('data/Planning_Districts.geojson')

districts_gdf = districts_gdf.to_crs(epsg=4326)

# plot the heatmap with origin
from folium.plugins import HeatMap

heatmap_data = walk[['origin_lat', 'origin_lon']].dropna().values.tolist()

o_des = folium.Map(location=[average_lat, average_lon], zoom_start=12, tiles='CartoDB positron')

HeatMap(heatmap_data).add_to(o_des)

for _, row in districts_gdf.iterrows():
    sim_geo = gpd.GeoSeries(row['geometry']).simplify(tolerance=0.001)
    geo_j = sim_geo.to_json()
    geo_j = folium.GeoJson(data=geo_j,
                           style_function=lambda x: {'fillColor': 'transparent', 'color': 'black'})
    geo_j.add_to(o_des)

o_des

Make this Notebook Trusted to load map: File -> Trust Notebook

# plot the heatmap with origin
from folium.plugins import HeatMap

heatmap_data2 = walk[['destination_lat', 'destination_lon']].dropna().values.tolist()

d_des = folium.Map(location=[average_lat, average_lon], zoom_start=12, tiles='CartoDB positron')

HeatMap(heatmap_data2).add_to(d_des)

for _, row in districts_gdf.iterrows():
    sim_geo = gpd.GeoSeries(row['geometry']).simplify(tolerance=0.001)
    geo_j = sim_geo.to_json()
    geo_j = folium.GeoJson(data=geo_j,
                           style_function=lambda x: {'fillColor': 'transparent', 'color': 'black'})
    geo_j.add_to(d_des)

d_des

Make this Notebook Trusted to load map: File -> Trust Notebook

from folium.plugins import HeatMap

# Assuming walk['origin_lat'] and walk['origin_lon'] are the coordinates for origins
heatmap_data_origins = walk[['origin_lat', 'origin_lon']].dropna().values.tolist()

# Assuming walk['destination_lat'] and walk['destination_lon'] are the coordinates for destinations
heatmap_data_destinations = walk[['destination_lat', 'destination_lon']].dropna().values.tolist()

# Create a Folium map centered around the average coordinates
heat_combined = folium.Map(location=[average_lat, average_lon], zoom_start=12, tiles='CartoDB positron')

# Add the origin heatmap layer with one color gradient
HeatMap(heatmap_data_origins, radius=7, blur=15, gradient={0.2: 'blue', 0.4: 'cyan', 0.6: 'lime', 0.8: 'yellow'}).add_to(heat_combined)

# Add the destination heatmap layer with another color gradient
HeatMap(heatmap_data_destinations, radius=7, blur=15, gradient={0.2: 'purple', 0.4: 'violet', 0.6: 'magenta', 0.8: 'pink'}).add_to(heat_combined)

# Overlay the district boundaries
for _, row in districts_gdf.iterrows():
    sim_geo = gpd.GeoSeries(row['geometry']).simplify(tolerance=0.001)
    geo_j = sim_geo.to_json()
    geo_j = folium.GeoJson(data=geo_j,
                           style_function=lambda x: {'fillColor': 'transparent', 'color': 'black', 'weight': 1})
    geo_j.add_to(heat_combined)
    

heat_combined

Make this Notebook Trusted to load map: File -> Trust Notebook

3. Explore Center City as the Pedestrain Busy District

3.1 Why and Who Walk in Center City

After uncovering the pedestrain busy district of center city, here in this section, I specifically examined the walk trips and pedestrains in center city.

In my exploration of pedestrian activities within Philadelphia’s central district, I identified distinctive patterns of movement. The data indicated a high frequency of ‘transfer’ activities, which suggests that the central district serves not only as a destination but also as a pivotal hub for pedestrian movement through the city. This is reflective of the dense network of public transit and the district’s role as a connector for various parts of the city. Among other prominent activities, ‘work for pay’ and ‘eat out (restaurant, drive-thru, etc.)’ were also significant, highlighting the central district’s commercial and social importance. Surprisingly, ‘home activities’ were the most frequent type of walk, which may be attributed to the residential pockets within the central district or the tendency of individuals to categorize errands close to home as ‘home activities’.

import geopandas as gpd
from shapely.geometry import Point

# filter out the walk trip in the central district

walk_gdf = gpd.GeoDataFrame(walk, geometry=gpd.points_from_xy(walk.origin_lon, walk.origin_lat))

walk_gdf.crs = districts_gdf.crs

central_polygon = districts_gdf.loc[districts_gdf['DIST_NAME'] == 'Central', 'geometry'].squeeze()

walk_in_central = walk_gdf[walk_gdf.geometry.within(central_polygon)]

walk_in_central.head()

	ID	HH_ID	PERSON_NUM	PERSON_ID	HH_WEIGHT	P_WEIGHT	TOUR_NUM	WKSUB_NUM	WKSUB_ID	TRIP_NUM	...	AGECAT	RACE	EDUCA	origin_tra	origin_lat	origin_lon	destination_tra	destination_lat	destination_lon	geometry
0	13	100206	1	10020601	204.106167	281.143358	1.0	NaN	1.002060e+11	3	...	45 to 54	NaN	Bachelor's or undergraduate degree	000804	39.948566	-75.167755	000300	39.957150	-75.171232	POINT (-75.16775 39.94857)
1	14	100206	1	10020601	204.106167	281.143358	1.0	NaN	1.002060e+11	4	...	45 to 54	NaN	Bachelor's or undergraduate degree	000300	39.957150	-75.171232	000804	39.948566	-75.167755	POINT (-75.17123 39.95715)
2	314	102372	1	10237201	209.139870	218.215946	1.0	NaN	1.023720e+11	4	...	35 to 44	NaN	Graduate degree	001002	39.945006	-75.146618	001002	39.945006	-75.146618	POINT (-75.14662 39.94501)
3	315	102372	1	10237201	209.139870	218.215946	1.0	NaN	1.023720e+11	5	...	35 to 44	NaN	Graduate degree	001002	39.945006	-75.146618	000902	39.947161	-75.156637	POINT (-75.14662 39.94501)
4	1009	109412	1	10941201	193.156252	82.416676	1.0	NaN	1.094120e+11	3	...	65 to 74	NaN	Graduate degree	000500	39.951955	-75.158176	000600	39.949386	-75.158334	POINT (-75.15818 39.95195)

5 rows × 95 columns

activity_mapping = { 1: 'Home activities not related to work, school, or online',
                 2: 'Homework, class related assignments or attended an online course',
                 3: 'Attended classes',
                 4: 'Attended other school activities (performances, meetings, clubs)',
                 5: 'Work for pay',
                 6: 'Personal business (banking or ATM, salon, library)',
                 7: 'Online personal business (banking, e-mail, etc.)',
                 8: 'Everyday shopping (grocery, drug store, gas, etc.)',
                 9: 'Major shopping (appliances, cars, home furnishings, clothes, etc.)',
                 10: 'Online shopping for products, services or goods',
                 11: 'Eat out (restaurant, drive-thru, etc.)',
                 12: 'Social (visit friends, relatives, etc.)',
                 13: 'Social/community/religious (meetings, worhship, wedding, funeral, etc.)',
                 14: 'Recreation- active participation (sports, exercise, walk the dog, etc.)',
                 15: 'Recreation-watch/observe (movies, concert, sports event, etc.)',
                 16: 'Medical (medical appointment, medical procedure, etc.)',
                 17: 'Pick up passenger',
                 18: 'Drop off passenger',
                 19: 'Change type of transportation/transfer',
                 20: 'Accompany household member',
                 96: 'Other activity'}

walk_in_central['ACTIV1'] = walk_in_central['ACTIV1'].map(activity_mapping)

# Plotting
plt.figure(figsize=(10, 8)) 
activity_counts = walk_in_central['ACTIV1'].value_counts()


ordered_activities = activity_counts.reindex(activity_mapping.values())


ordered_activities.plot(kind='bar')

plt.title('Frequency of Walk Activity Type')
plt.xlabel('Activity')
plt.ylabel('Count')


plt.xticks(rotation=90) 


for index, value in enumerate(ordered_activities.values):
    plt.text(index, value, f'{value}') 

plt.tight_layout() 
plt.show()

D:\Anaconda\Install\envs\musa-550-fall-2023\lib\site-packages\geopandas\geodataframe.py:1538: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy
  super().__setitem__(key, value)

Upon reprofiling the pedestrian demographic within the central district, I found that the socioeconomic characteristics closely mirrored those observed citywide. Vehicle ownership, income, age, race, gender, and education level demonstrated similar trends, with no significant deviations. This lack of disparity suggests a consistent pedestrian demographic across the city, irrespective of the more intense walking activity in the central district. Most walkers belonged to households without a vehicle, spanned across a diverse range of income levels, and were predominantly White/Caucasian with at least a graduate degree, reflecting the broader city’s pedestrian profile. However, it was interesting to note that the central district’s pedestrian population was just as diverse as the rest of Philadelphia, reinforcing the notion that walking as a mode of transport transcends socioeconomic boundaries within urban centers.

fig, axs = plt.subplots(2, 3, figsize=(15, 10))  

walk_in_central['TOT_VEH'].value_counts().sort_index().plot(kind='bar', ax=axs[0, 0], title='Total Vehicles')
walk_in_central['INCOME'].value_counts().sort_index().plot(kind='bar', ax=axs[0, 1], title='Income')
walk_in_central['GEND'].value_counts().sort_index().plot(kind='bar', ax=axs[0, 2], title='Gender')
walk_in_central['AGECAT'].value_counts().sort_index().plot(kind='bar', ax=axs[1, 0], title='Age Category')
walk_in_central['RACE'].value_counts().sort_index().plot(kind='bar', ax=axs[1, 1], title='Race')
walk_in_central['EDUCA'].value_counts().sort_index().plot(kind='bar', ax=axs[1, 2], title='Education')

plt.tight_layout()
plt.show()

3.2 What are There in Center City Attracting People to Walk Here

In the latter half of my analysis of Center City, I delved into the urban elements that contribute to its high pedestrian traffic. I used OpenStreetMap to overlay various components of the urban environment, revealing a dense and intricate web of walkways that form the backbone of pedestrian movement in the area. My mapping highlighted an extensive network of roads frequented by pedestrians, which are well-serviced by an array of public transit options, including numerous bus stops and metro stations strategically located throughout the district.

The presence of these amenities suggests a cityscape thoughtfully designed to encourage walking, where the convenience of public transport complements pedestrian pathways. Moreover, my data visualizations pinpointed clusters of restaurants, schools, and commercial blocks, which serve as vital nodes of activity, drawing people onto the streets. These commercial hubs are not only destinations but also waypoints that enliven the pedestrian experience with their vibrancy and accessibility.

import osmnx as ox
ox.config(use_cache=True, log_console=True)

# For bus stops
bus_stops = ox.geometries_from_place('Center City,Philadelphia, PA', tags={'highway': 'bus_stop'})

# For metro stops 
metro_stops = ox.geometries_from_place('Center City,Philadelphia, PA', tags={'railway': 'station'})

# For restaurants
restaurants = ox.geometries_from_place('Center City,Philadelphia, PA', tags={"amenity": ["pub", "bar", "restaurant"]})

# For schools
schools = ox.geometries_from_place('Center City,Philadelphia, PA', tags={'amenity': 'school'})

# For commercial blocks
commercial = ox.geometries_from_place('Center City,Philadelphia, PA', tags={'landuse': 'commercial'})

# For parks
parks = ox.geometries_from_place('Center City,Philadelphia, PA', tags={'leisure': 'park'})

C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\2733245946.py:2: UserWarning: The `utils.config` function is deprecated and will be removed in a future release. Instead, use the `settings` module directly to configure a global setting's value. For example, `ox.settings.log_console=True`.
  ox.config(use_cache=True, log_console=True)
C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\2733245946.py:5: UserWarning: The `geometries` module and `geometries_from_X` functions have been renamed the `features` module and `features_from_X` functions. Use these instead. The `geometries` module and function names are deprecated and will be removed in a future release.
  bus_stops = ox.geometries_from_place('Center City,Philadelphia, PA', tags={'highway': 'bus_stop'})
C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\2733245946.py:8: UserWarning: The `geometries` module and `geometries_from_X` functions have been renamed the `features` module and `features_from_X` functions. Use these instead. The `geometries` module and function names are deprecated and will be removed in a future release.
  metro_stops = ox.geometries_from_place('Center City,Philadelphia, PA', tags={'railway': 'station'})
C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\2733245946.py:11: UserWarning: The `geometries` module and `geometries_from_X` functions have been renamed the `features` module and `features_from_X` functions. Use these instead. The `geometries` module and function names are deprecated and will be removed in a future release.
  restaurants = ox.geometries_from_place('Center City,Philadelphia, PA', tags={"amenity": ["pub", "bar", "restaurant"]})
C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\2733245946.py:14: UserWarning: The `geometries` module and `geometries_from_X` functions have been renamed the `features` module and `features_from_X` functions. Use these instead. The `geometries` module and function names are deprecated and will be removed in a future release.
  schools = ox.geometries_from_place('Center City,Philadelphia, PA', tags={'amenity': 'school'})
C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\2733245946.py:17: UserWarning: The `geometries` module and `geometries_from_X` functions have been renamed the `features` module and `features_from_X` functions. Use these instead. The `geometries` module and function names are deprecated and will be removed in a future release.
  commercial = ox.geometries_from_place('Center City,Philadelphia, PA', tags={'landuse': 'commercial'})
C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\2733245946.py:20: UserWarning: The `geometries` module and `geometries_from_X` functions have been renamed the `features` module and `features_from_X` functions. Use these instead. The `geometries` module and function names are deprecated and will be removed in a future release.
  parks = ox.geometries_from_place('Center City,Philadelphia, PA', tags={'leisure': 'park'})

# the "walk" network
G = ox.graph_from_address("Center City, Philadelphia, PA", network_type="walk")
ox.plot_graph(ox.project_graph(G), node_size=0)

(<Figure size 800x800 with 1 Axes>, <Axes: >)

G_cc = ox.graph_from_polygon(central_polygon, network_type="walk")

# only get the edges
cc_edges = ox.graph_to_gdfs(G_cc, edges=True, nodes=False)

# add the neighborhood boundaries
boundary = gpd.GeoSeries([central_polygon], crs="EPSG:4326")

cc_edges.explore(tiles='cartodb positron')

Make this Notebook Trusted to load map: File -> Trust Notebook

import geopandas as gpd

centroid = central_polygon.centroid
central_point = centroid

longitude = central_point.x
latitude = central_point.y

import folium

bus = folium.Map(location=[latitude, longitude], zoom_start=13)

for idx, row in bus_stops.iterrows():
    folium.Marker(location=[row.geometry.y, row.geometry.x], popup='Bus Stop').add_to(bus)

bus

Make this Notebook Trusted to load map: File -> Trust Notebook

metro = folium.Map(location=[latitude, longitude], zoom_start=13)

for idx, row in metro_stops.iterrows():
    folium.Marker(location=[row.geometry.y, row.geometry.x], popup='Metro Stops').add_to(metro)

metro

Make this Notebook Trusted to load map: File -> Trust Notebook

restaurant_points = restaurants[restaurants.geometry.type == 'Point']

map_center = restaurant_points.geometry.unary_union.centroid

res_m = folium.Map(location=[map_center.y, map_center.x], zoom_start=14)

for idx, row in restaurant_points.iterrows():
    folium.Marker(
        location=[row.geometry.y, row.geometry.x], 
        popup='Restaurant'
    ).add_to(res_m)

res_m

Make this Notebook Trusted to load map: File -> Trust Notebook

schools_points = schools[schools.geometry.type == 'Point']

map_center = schools_points.geometry.unary_union.centroid

schools_map = folium.Map(location=[map_center.y, map_center.x], zoom_start=14)

for idx, row in schools_points.iterrows():
    marker = folium.Marker(location=[row.geometry.y, row.geometry.x], popup='Schools')
    marker.add_to(schools_map)

schools_map

Make this Notebook Trusted to load map: File -> Trust Notebook

from shapely.geometry import Polygon, MultiPolygon
commerce_m = folium.Map(location=[latitude, longitude], zoom_start=13)

for idx, row in commercial.iterrows():
    if isinstance(row.geometry, (Polygon, MultiPolygon)):
        folium.GeoJson(row.geometry, 
                        popup='Commercial Block',
                        style_function=lambda x: {'fillColor': 'blue', 'color': 'blue'}
                       ).add_to(commerce_m)

commerce_m

Make this Notebook Trusted to load map: File -> Trust Notebook

In addition to these urban features, the green spaces represented by parks emerged as significant attractors for walkers, offering respite and recreational space amidst the urban fabric. These verdant areas likely contribute to the appeal of walking by enhancing the environmental quality and aesthetic of the city.

parks_map = folium.Map(location=[latitude, longitude], zoom_start=13)

for idx, park in parks.iterrows():
    if isinstance(park.geometry, (Polygon, MultiPolygon)):
        folium.GeoJson(
            data=park.geometry.__geo_interface__,
            popup='Park',
            style_function=lambda x: {'fillColor': 'green', 'color': 'green'}
        ).add_to(parks_map)

parks_map

Make this Notebook Trusted to load map: File -> Trust Notebook

Overall, my investigation paints a picture of Center City as a robust hub of pedestrian activity, supported by a well-integrated system of walkways, transit stops, commercial establishments, educational institutions, and green spaces. Each element plays a role in creating an inviting pedestrian environment, showcasing the multifaceted allure of Center City for walkers. The harmonious interplay between these factors not only facilitates but also enriches the pedestrian experience, affirming Center City’s reputation as a vibrant and walkable urban center.

4. Why Center City Become a Pedestrain Busy Zone

To further understand what drives pedestrian movement in Center City, I conducted a regression analysis with the aim of quantifying the influence of various urban features on walking trip counts. Using Python’s scikit-learn library, I fitted a linear regression model with variables including the count of bus stops, metro stops, restaurants, schools, commercial areas, and park areas.

import geopandas as gpd
from shapely.geometry import Point
from sklearn.linear_model import LinearRegression
import pandas as pd

districts_gdf.head()
central = districts_gdf.loc[districts_gdf['DIST_NAME'] == 'Central', 'geometry']

philly_tract.set_crs(epsg=4326, inplace=True)

central_gdf = gpd.GeoDataFrame(central)

central_gdf.set_crs(philly_tract.crs, inplace=True)

central_tracts = gpd.sjoin(philly_tract, central_gdf, how="inner", op='intersects')

D:\Anaconda\Install\envs\musa-550-fall-2023\lib\site-packages\IPython\core\interactiveshell.py:3448: FutureWarning: The `op` parameter is deprecated and will be removed in a future release. Please use the `predicate` parameter instead.
  if await self.run_code(code, result, async_=asy):

if 'index_right' in central_tracts.columns:
    central_tracts = central_tracts.drop(columns=['index_right'])

import geopandas as gpd

def count_pois(geo_df, tracts_df, poi_name):
    join = gpd.sjoin(geo_df, tracts_df, predicate='within')
    count = join.groupby('index_right').size()
    tracts_df[poi_name] = count.reindex(tracts_df.index, fill_value=0)

def calculate_area(geo_df, tracts_df, area_name):
    join = gpd.sjoin(geo_df, tracts_df, predicate='intersects')
    area = join.groupby('index_right').apply(lambda g: g.geometry.area.sum())
    tracts_df[area_name] = area.reindex(tracts_df.index, fill_value=0)

count_pois(bus_stops, central_tracts, 'bus_stop_count')
count_pois(metro_stops, central_tracts, 'metro_stop_count')
count_pois(restaurant_points, central_tracts, 'restaurant_count')
count_pois(schools_points, central_tracts, 'school_count')
count_pois(walk_in_central, central_tracts, 'walktrip_count')


calculate_area(commercial, central_tracts, 'commercial_area')
calculate_area(parks, central_tracts, 'park_area')

central_tracts.reset_index(drop=True, inplace=True)

C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\1808947791.py:10: UserWarning: Geometry is in a geographic CRS. Results from 'area' are likely incorrect. Use 'GeoSeries.to_crs()' to re-project geometries to a projected CRS before this operation.

  area = join.groupby('index_right').apply(lambda g: g.geometry.area.sum())
C:\Users\gaoxi\AppData\Local\Temp\ipykernel_29064\1808947791.py:10: UserWarning: Geometry is in a geographic CRS. Results from 'area' are likely incorrect. Use 'GeoSeries.to_crs()' to re-project geometries to a projected CRS before this operation.

  area = join.groupby('index_right').apply(lambda g: g.geometry.area.sum())

central_tracts.head()

	STATEFP	COUNTYFP	TRACTCE	GEOID	NAME	NAMELSAD	MTFCC	FUNCSTAT	ALAND	AWATER	...	centroid	centroid_lat	centroid_lon	metro_stop_count	restaurant_count	school_count	commercial_area	park_area	bus_stop_count	walktrip_count
0	42	101	980000	42101980000	9800	Census Tract 9800	G5020	S	8923895	1501892	...	POINT (-75.20022 39.98506)	39.985064	-75.200217	0	0	0	0.000000e+00	0.000310	1	0
1	42	101	036700	42101036700	367	Census Tract 367	G5020	S	801972	0	...	POINT (-75.14399 39.96073)	39.960726	-75.143987	1	2	0	1.814250e-06	0.000003	0	18
2	42	101	037600	42101037600	376	Census Tract 376	G5020	S	717301	0	...	POINT (-75.15516 39.95990)	39.959902	-75.155163	0	4	4	2.725569e-07	0.000001	0	7
3	42	101	036900	42101036900	369	Census Tract 369	G5020	S	1859160	193283	...	POINT (-75.19370 39.94910)	39.949097	-75.193703	0	0	0	0.000000e+00	0.000000	0	0
4	42	101	036600	42101036600	366	Census Tract 366	G5020	S	1004315	1426276	...	POINT (-75.13895 39.94486)	39.944863	-75.138949	0	3	0	0.000000e+00	0.000008	1	16

5 rows × 23 columns

The correlation heatmap provided initial insights into the relationships between these variables, showing a strong positive correlation between the number of restaurants and walking trips, which aligns with the idea that dining options significantly contribute to pedestrian traffic. A moderate positive correlation was also observed between the number of bus and metro stops and walking trips, supporting the notion that public transit accessibility encourages walking.

train_set = central_tracts[['bus_stop_count', 'metro_stop_count', 'restaurant_count', 'school_count', 'commercial_area', 'park_area']]

sns.heatmap(
    train_set.corr(),
    cmap="coolwarm",
    annot=True,
    vmin=-1,
    vmax=1
)

<Axes: >

from sklearn.linear_model import LinearRegression

X = central_tracts[['metro_stop_count', 'restaurant_count', 'school_count', 'commercial_area', 'park_area']]
y = central_tracts['walktrip_count']  

model = LinearRegression()
model.fit(X, y)

LinearRegression()

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Upon fitting the linear regression model, the coefficients revealed the relative impact of each feature. The presence of restaurants had a noticeable positive coefficient, suggesting a strong relationship with pedestrian activity. Interestingly, the number of schools had a negative coefficient, which could indicate that the areas around schools in Center City do not necessarily generate additional foot traffic, or other factors may be at play that weren’t captured in the model.

The model’s mean squared error (MSE) and coefficient of determination (R²) were evaluated to measure its performance. With an R² value of approximately 0.426, the model explains around 42.6% of the variance in walking trips. While this indicates that the model has captured a significant portion of the factors affecting pedestrian movement, there is still a substantial amount of variance unexplained, which could be due to unmodeled variables or the complexity of pedestrian dynamics that are not easily captured by linear relationships.

print(f'Coefficients: {model.coef_}')
print(f'Intercept: {model.intercept_}')

Coefficients: [ 9.72257383e+00  2.25749357e+01 -1.44216056e+00  1.83764737e+00
 -8.41207906e+05  1.29592761e+04]
Intercept: 20.57387711612844

from sklearn.metrics import mean_squared_error, r2_score

y_pred = model.predict(X)

print(f'Mean squared error (MSE): {mean_squared_error(y, y_pred)}')
print(f'Coefficient of determination (R^2): {r2_score(y, y_pred)}')

Mean squared error (MSE): 1977.6339251018762
Coefficient of determination (R^2): 0.42607554891522836

In conclusion, my regression analysis has shed light on some key factors that appear to influence walking trip counts in Center City. While the findings confirm the importance of amenities like restaurants and public transit stops, they also prompt further investigation into other potential influences on pedestrian activity. This model serves as a foundational step towards a more nuanced understanding of urban walkability and the myriad elements that foster an inviting pedestrian environment.

5. Conclusion & Next Step

The findings from this project underscore the intricate relationship between urban design and pedestrian behavior. Center City’s robust network of walkways, transit stops, and commercial and recreational areas not only accommodates but also encourages walking as a primary mode of transportation. My regression analysis, while highlighting the substantial impact of dining options and public transit, also points to the complexity of pedestrian dynamics and the potential for further research. With an R² value reflecting that only a portion of walking trip variance is explained, there’s room to explore additional variables and more sophisticated models. This project has laid the groundwork for a deeper exploration into urban walkability, offering insights that could inform city planners and policymakers seeking to enhance pedestrian experiences in urban landscapes. The conclusion we draw is that while pedestrian movement is influenced by a myriad of factors, the centrality of accessible amenities and a pedestrian-friendly infrastructure cannot be overstated in fostering a vibrant, walkable city. In the next step, I would like to further explore the pedestrain desert, making comparison with the pedestrain busy zone and find out the back reason of lacking of pedestrains in those areas.